STTELLA
The problem
Where companies lose time
People retype data from invoices and documents, look for information in several places and assemble the same reports by hand. Much of that work can be automated.
- Retyping data from invoices, orders and forms
- Searching documentation, e-mails and files for answers
- Building recurring reports manually
- Giving customers and staff the same answers over and over
- Reviewing documents by reading everything one by one
The solution
What we can build
Every project starts with the process, not the model. We match AI to the task — and where plain automation is enough, we don't add AI for its own sake.
- AI assistants for employees
- Smart search across company knowledge
- OCR and automatic data extraction from documents
- Invoice and document processing
- Automatic reports and e-mail notifications
- Document generation
- AI Vision: reading labels, markings and numbers from images
- Decision-support systems
Example uses
What it can look like in practice
Example scenarios — the actual scope is agreed after a conversation and a look at your data.
- Admin: invoice data goes into the system without retyping
- Customer service: an assistant suggests answers based on company documentation
- Production: reading markings from cameras and alerting when something is off
- Management: a recurring report built from company data and sent by e-mail
- HR: quick search across policies and procedures
Data
Where the model runs
Part of our AI processing runs locally on our own GPUs — the data does not have to leave our infrastructure. Where processing happens, and which data may reach a model, is agreed with the client before deployment.
We don't promise 100% accuracy and we don't replace people — the system supports your team and flags what needs attention.
Honestly
When you don't need AI
If a process covers a handful of documents a month or changes every week, simpler automation without AI is often cheaper and more reliable. We will tell you so during the first conversation.
Process
How we work together
- Conversation and scope we learn the process and how long it takes today
- Data review we look at sample documents and data
- Prototype a first version runs on your examples
- Deployment integration with the systems you already use
- Improvement adjustments based on real use
Technology
What we use
- Python
- Local language models on our own GPUs
- OCR and document analysis
- PostgreSQL and SQLite databases
- API integrations
- Automatic e-mail reports
What it costs
Automating a small process — from PLN 290. Process automation — typically from a few thousand PLN. AI, OCR and document analysis — scoped after a review of your data.
Work
Examples from our projects
- Agent Pracy — AI drafts cover letters and portfolios, and job matching uses a transparent, itemised score.
FAQ
Frequently asked questions
Will company data be sent to an external AI service?
It doesn't have to be. Part of our AI processing runs locally on our own GPUs. Where processing happens is agreed with you before deployment.
How much does AI automation cost?
Automating a small process starts at PLN 290; process automation typically from a few thousand PLN. AI, OCR and document-analysis projects are priced after a review of your data.
Will AI replace employees?
That is not the goal. The system supports the team: it takes over repetitive steps and flags what needs a person's attention.
Will it work with the software we already use?
We work with existing systems and build integrations. We check how data can be exchanged with your software.
Where do we start?
Describe the process that takes the most time. After a short conversation we prepare a scope and a quote.